The Formation of Biphenyls from Derivatives of Benzene-1, 4

Dana R. Reed, Michael C. Hare, Alireza Fattahi, Gyusung Chung, Mark S. Gordon, ... Duncan M. A. Grieve , Graham E. Lewis , Michael D. Ravenscroft , Pe...
0 downloads 0 Views 769KB Size
MICHAELJ. S. DEWARAND K. NARAYANASWAMI

2422 [CONTRIBUTION FROM

THE

Vol 86

GEORGE HERBERT JOXES LABORATORY, UNIVERSITY OF CHICAGO, CHICAGO, ILL.]

The Formation of Biphenyls from Derivatives of Benzene-l,4-diazooxide. Substitution

Electrophilic

B Y MICHAEL J . s. DEWAR? 4 K D K . NAR.4YANASWAMI RECEIVED JAXCARY 16, 1964 Earlier work3 had shown that the thermal decomposition of derivatives of benzene- 1,4-diazooxides in benzene, or in derivatives of benzene, in the presence of catalytic amounts of alcohol gives derivatives of 4-hydroxybiphenyl in good yield. We have re-examined this reaction in detail and have shown that, contrary to an earlier sugg e ~ t i o nit, ~does not involve a free radical niechanis~n. Decomposition of the diazooxide gives a carbene which attacks the aromatic substrate. The process is essentially an electrophilic substitution, the first step in which is the formation of a x-complex or spiran. The bearing of this on the mechanism of electrophilic aromatic substitution is discussed.

Previous work3 had shown t h a t the thermal decomposition of 3,5-dibromobenzene-1,4-diazooxide (Ia) in chlorobenzene, 0- or m-dichlorobenzene, or bromobenzene, gives polymers together with free bromine, and that the polymers are copolymers of the repeating units I1 and 111. This was interpreted in terms of a mechanism in which I a decomposes to the biradical IV. One of the unpaired electrons in IV occupies a a-MO similar to that occupied by the odd electron in a phenoxy radical, while the other occupies an in-plane u-type A0 which is not mesomerically stabilized. The unstabilized radical center should show the same reactivity as t h a t in a free aryl radical; attack on the solvent accounts for the incorporation of solvent molecules in the polymer. The second radical center should be much less reactive; however, i t can apparently displace a bromine atom from some other molecule, forming an oxide bridge together with free bromine. The feasibility of this process was established by experiments with bromophenols. Thus oxidation of 2,4,6tribromophenol gave a polymer together with free bromine, the polymer presumably being a polyphenylene oxide formed by an attack of intermediate aryloxy radicals on the tribromophenol. Various polymers of this kind were prepared in this way from other phenol derivatives, and Staffin and Price4 have also independently prepared such materials by an analogous route.

x+X Ni+

Ia, X=Br b, X=CI

benzonitrile. In these cases little or no polymer was formed, no bromine was evident, and the product was a mixture of isomeric 4hydroxybiphenyls (V), the yields being excellent. I t was then found that similar biphenyl derivatives could also be obtained from the chlorinated or brominated benzenes if the reaction were carried out in presence of a little alcohol; as little as 1% of methanol or ethanol was sufficient to repress the polymerization completely, and to give a biphenyl derivative in excellent yield. Competitive experiments showed t h a t different benzene derivatives reacted a t very similar rates-with Ia in the presence of alcohols, seeming to suggest that the biaryl synthesis also involved a free-radical mechanism. I t was therefore postulated that both the polymerization and the biaryl synthesis involve a prelirninary attack by IV on the solvent, the resulting biradical VI either stabilizing itself by a hydrogen transfer giving V or undergoing conversion to polymers I t was suggested t h a t the remarkable effect of alcohols might be due to their acting as hydrogen transfer agents zda intermediate radicals such as .CHzOH. The orientation of the products from chlorobenzene indicated that the attack was electrophilic in nature, the product being a mixture of 2‘- and 4’-chloro-2,6-dibroino-lhydroxybiphenyls (V, Xr = 2- or 4chlorophenyl) ; however, this observation did not in itself exclude a radical mechanism for the reaction since it has been shown5that phenyl radicals carrying electron-withdrawing substituents behave as electrophilic reagents. Polarization of the a-electrons in IV should make it approximate to V I I ; one might therefore expect I V to show a strong electrophilic tendency when it acts as an arylating agent. 0.

0..*-

VI1 VI

Entirely different results were obtained when conipound I was heated in benzene, or in anisole, fluorobenzene, N,N-dimethylaniline, methyl benzoate, or (1) T h i s work was supported by the Office of Naval Research through Contract Xonr 2121-21 and by a grant from t h e National Science Foundation. ( 2 ) Correspondence should be addressed t o Department of Chemistry, T h e University of Texas, Austin, Texas. (3) h l J S Dewar and A . K.James, J Chem Soc , 917, 1265 (1958). ( 3 ) G D Staffin and C C. Price, J A m . Chem S o c . , 82, 3632 (19601

This biaryl synthesis is clearly of interest both as a possible practical route to biaryls and also from a theoretical standpoint. If it does indeed involve a free radical substitution process, i t would provide an excellent system for studying orientation and reactivity in such reactions; for no other’radicalsubstitution reaction has ( 5 ) J I . G. Cadogan, D H H e y , and G . H Williams, .I (1955).

Chr?n S o c , 1425

June 20, 1964

BIPHENYLS FROM BENZENE-~,~-DIAZOOXIDE DERIVATIVES

given such good yields of product. We have therefore re-examined the reactions of diazooxides with benzene derivatives in some detail. Our work has been carried out mostly with 3,5-dichlorobenzene-1,4-diazooxide (Ib) which has been shown3 to react in much the same way as the dibromide I a . Decomposition of I in a benzene derivative CsH6X can give three biaryl derivatives (V) in which the substituent X occupies an 0-,m-, or p-position. Our object was to determine the partial rate factors for attack in each position of C6H6X. For this we needed to know the rates of over-all reaction of C6H5X relative to benzene, and the proportions of the three isomers in the product; in order to do this we had to synthesize the various biphenyl derivatives independently since most of them were a t that time unknown.

Experimental T h e reactions of I were carried out in an excess of the benzene derivative as solvent a t 70'; the substrates used were benzene, fluorobenzene, chlorobenzene, bromobenzene, anisole, methyl benzoate, and benzonitrile. U'hen evolution of nitrogen had ceased (usually after 6 h r . ) the excess solvent was steam distilled and the biphenyl derivative V isolated with alkali and methylated with diazomethane in ether. Preliminary work had shown t h a t the best way of analyzing the product was by vapor phase chromatography of the methyl ethers, and t h a t the methylation of V to t h e ether was quantitative under t h e conditions we used. Chromatography on silicone gum rubber a t 300" very easily separated compounds derived from different benzene derivatives, and it also separated the o-isomer from the m- and p-isomers in each case. The inseparable mixtures of m- and p-isomers were analyzed by infrared spectroscopy, using the spectra of the pure compounds which we prepared independently. T h e reactivities of the benzene derivatives relative t o benzene were determined by a competitive method, using a mixture of benzene and CsHjX as solvent. Yearly all our work was carried out with the dichlorodiazooxide I b since the biphenyl derivatives V derived from Ia tended t o decompose during chromatography. We also looked for a possible isotope effect in the reaction of I b with 1,3,5-trideuteriobenzene; in this case the product was analyzed by mass spectrometry.6 The role of alcohol in the reaction was studied by using a,adideuterioethanol as catalyst; the fate of the deuterium was established by infrared and n.m.r. spectroscopy. Synthesis of Reference Compounds.-We required a series of ( V I I I ) containX'-substituted 3,5-dichloro-4-hydroxybiphenyls ing fluorine, chlorine, bromine, methoxyl, methoxycarbonyl, or cyano in each of the three possible positions (2', 3', 4 ' ) . Most of these compounds were prepared either by a n Cllmann reaction of an X-substituted iodobenzene and p-iodoanisole to form an X'-substituted 4-methoxybiphenyl ( I X ) , which was then chlorinated to 1.111, or directly by a n Cllmann reaction of the appropriate iodobenzene with 2,6-dichloro-4-iodoanisole( X ) . Full details are given below. n r

c1

X

V I I I , X = F , C1, Br, MeO, MeOOC, C S , I X , X as in VI11

3,5-Dibromo-2 '-chloro-4-hydroxybiphenyl.-Bromine (0.4 ml.) in acetic acid (5 ml.) was added t o a solution of 2-chloro-4'hydroxybiphenyl (@.71g . ) in acetic acid (3 ml.). Precipitation with water and recrystallization from petroleum ether (b.p, ( 6 ) We a r e very grateful t o Dr S. Meyerson of t h e American Oil Co for carrying out these determinations.

2423

30-35") gave the dibromochlorohydroxybiphenyl ( 1.50 g., 79yc) as needles, m.p. 132-134" (lit.3 124-126'). 3,5-Dibromo-4'-chloro-4-hydroxybiphenyl was prepared likewise from 4'-chloro-4-hydroxybiphenyl in 75% yield, m.p. 166167" (lit.3167.5-168.5D). 3,4',5-Trichloro-4-methoxybiphenyl.-The calculated amount of chlorine was passed into a solution of 4-chloro-4'-hydroxybiphenyl (0.68 9.) in acetic acid ( 7 ml.) and carbon tetrachloride (12 ml.) a t room temperature. Excess chlorine was removed with a current of air and the solution evaporated under vacuum. crystallized The residue of 3,4',5-trichloro-4-hydroxybiphenyl from petroleum ether (b.p. 60-65") in white needles (0.756 g., go%,), m . p . 118-149'. Methylation with diazomethane in ether gave 3,4',5-trichloro-4-methoxybiphenylin almost quantitative yield, m . p . 98-99", Anal. Calcd. for C13H90C13: C , 54.29; H , 3.15; C1, 36.99. Found: C, 54.64; H , 3.35; C1, 37.23. 3,3 ',5-Trichloro-4-methoxybiphenylwas prepared likewise by chlorination of 3-chloro-4'-hydroxybiphenyl to 3,3',5'-trichloro4'-hydroxybiphenyl, m.p. 121-122', followed by methylation; m.p. 91-92", Anal. Found: C, 54.57; H , 3.24; C1, 36.98. 2,3',5'-Trichloro-4'-methoxybiphenyl was prepared likewise t'ia the phenol, m . p . 106-107'; m.p. 83-84', Anal. Found: C, 54.78; H , 3.21; C1, 36.96. Methyl 3',5 '-Dichloro-4'-methoxybiphenyl-4-carboxylate.--A mixture of 2,6-dichloro-4-iodoanisole (7.0 g.), methyl 4-iodobenzoate (5.0 g.), and copper bronze (4.5 g.) was heated 7 hr. a t 165". T h e residue was extracted with chloroform and the extract chromatographed on alumina from benzene-light petroleum ( 1 : l ) . T h e first and last fractions consisted, respectively, of methyl biphenyl-4,4'-dicarboxylate( 0 . 5 g., m.p. 214') and 3,3',5,5'-tetrachloro-4,4'-dimethoxybiphenyl( 3.0 g., m.p. 1 8 5 186' ). The middle fraction of methyl 3 ',5'-dichloro-4'-methoxybiphenyl4-carboxylate (1.0 g . ) had m.p. 132-133". Anal. Calcd. for CljH1203C12: C, 57.90; H , 3.89; C1, 22.79. Found: C, 58.01; H , 4.03; C1, 22.80. Methyl 3',5'-dic~oro-4'-methoxybiphenyl-3-carboxylatewas prepared likewise from 2,6-dichloro-4-iodoanisole (9.8 g.) and methyl 3-iodobenzoate (7.0 g . ) ; 2.6 g., m.p. 115-116'. A n a l . Found: C, 58.00; H , 3.86. Methyl 3',5'-dichloro-4'-methoxy-2-carboxylate was prepared likewise from 2,6-dichloro-4-iodoanisole (7.0 9 . ) and methyl 2-bromobenzoate (5.0 g . ) ; the ester (2.15 g . ) had m . p . 87-88'. Anal. Found: C , 58.02; H , 3 . 9 1 ; C1,22.67. 3,5-Dichloro-4'-fluoro-4-methoxybiphenylwas prepared likewise from p-fluoroiodobenzene (10.0 g.), 2,6-dichloro-4-iodoanisole (13.6 g.), and copper bronze (8.9 g.); 1.35 g., m.p. 94-95", Anal. Calcd. for CI3HsOC12F: C , 57.59; H , 3.36; CI, 26.16. Found: C , 57.95; H , 3.50; C1, 26.80. 3,5-Dichloro-3'-fluoro-4-methoxybiphenylwas prepared likewise; 1.57 g., m.p. 100-102". A n a l . Found: C , 57.25; H , 3.52; C1, 26.92. 3,5-Dichloro-2'-fluoro-4-methoxybiphenylwas prepared likewise; 0.87 g., m.p. 86-87". Anal. Found: C, 57.56; H , 3.22; CI, 26.42. 3,5-Dichloro-4,4'-dimethoxybiphenyl.-Copper bronze (9.0 g. ) was added over 20 min. to a stirred mixture of p-iodoanisole (12.0 9 . ) a t 195" and heat(7.0 g.) and 2,6-dichloro-4-iodoanisole ing continued for 12 h r , The dichlorodimethoxybiphenyl(2.51 g . ) was isolated a s before and after crystallization from methanol had m.p. 60-61'. A n a l . Calcd. for CI4H1?O2CI?:C , 59.38; H , 4.27; C1, 25.04. Found: C, 59.39; H , 4.46; C1, 25.08. 3,5-Dichloro-2',4-dimethoxybiphenyl was prepared likewise from 2,6-dichloro-4-iodoanisole (7.0 9 . ) and m-iodoanisole (5.0 g . ) ; 2.5 g., m.p. 66-67'. A n a l . Found: C, 59.39; H , 4.50; C1, 25.18. Nitration of 3,[email protected] of 3,5-dichloro-4-methoxybiphenylwith nitric acid in acetic anhydride gave a mixture of mononitro derivatives which was separated by fractional crystallization from methanol into 3,5-dichloro-4methoxy-2'-nitrobiphenyl, m.p. 101 ', and 3,5-dichloro-4-methoxy-4'-nitrobiphenyl, m . p . 155-156', the over-all yields being 32 and 46%, respectively. Anal. Calcd. for CI3H,SO3Cl2: C, 52.37; H , 3.04; N,4.70; CI, 23.79. Found for 2'-isomer: C, 52.75; H, 2.90; N , 1.76; CI, 23.60. Found for 4'4somer: C, 52.45; H , 3.15; S ,4.88; C1, 23.78.

2424

MICHAEL J. S. DEWARA N D K . NARAYANASWAMI

4-Amino-3',5'-dichloro-4'-methox~bi~hen~l.-Reduction of the nitro compound with iron filings in acetic acid gave the amine in almost theoretical yield, crystallizing from petroleum ether (b.p. 60-68') in needles, m p. 78". Anal. Calcd. for C13HllNOC12: C, 58.23; H , 4.14; N, 5.22; C1,26.45. Found: C, 58.60; H, 3.91; N, 5.47; CI, 26.51. 3,5-Dichloro-2 '-hydroxy-4-methoxybiphenyl.-Reduction of 3,5-dichloro-4-methoxy-2'-nitrobiphenylas above gave the crude amine as a viscous oil (1.65 9 . ) which was diazotized; the diazonium solution on pouring into boiling dilute sulfuric acid gave 3,5-dichloro-2 '-hydroxy-4-methoxybiphenyl which separated from petroleum ether ( b . p . 6C-68') in needles (3.5 g . , 75yc), m . p . 118-1 19'. Anal. Calcd. for C13Hl~02C12: C , 58.02; H , 3.75; C1, 26.35. Found: C, 58.07; H, 3.86; CI, 26.27. 3,5-Dichloro-2',4-dimethoxybiphenyl.-Methylation of the above hydroxybiphenyl (3.0 g . ) with dimethyl sulfate (3.0 9 . ) in aqueous potassium hydroxide (5 ml. of 20%) gave the ether which crystallized from petroleum ether (b.p. 30-35') in needles (3.0 g., 95%), m.p. 38". Anal. Calcd. for C I ~ H I Z O Z C LC, : 59.38; H , 4.27; CI, 25.04. Found: C, 59.41; H , 4.32; CI, 25.04. 3,5-Dichloro-2 '-cyano-4-methoxybiphenyl was prepared in the usual way from the corresponding amine by diazotization and treatment with cuprous cyanide; the nitrile after crystallization from petroleum ether ( b . p . 60-68") had m . p . 176-177'. Anal. Calcd. for C14HgSOC12: C, 60.46; H, 3.26; N,5.04; C1, 25.50. Found: C , 60.25; H , 3.34; K , 4.84; C1, 25.79. 3,5-Dichloro-4'-cyano-4-methoxybiphenylwas prepared likewise; m.p. 147-148'. Anal. Found: C , 61.00; H , 3.10; X , 4.90; CI, 25.49. 2-Bromo-3 ' , 5 '-dichloro-4 '-methoxybiphenyl was prepared in an analogous manner from the amine by diazotization and treatment with cuprous bromide; the bromodichlorodimethoxybiphenyl crystallized from petroleum ether in needles (yield 85%), m.p. 80-81 O . Anal. Calcd. for Cl3H90BrCI2: C , 47.02; H , 2.73; Br, 24.07; CI, 21.36. Found: C , 47.00; H , 2.70; Br, 24.05; C1, 21.20. 4-Bromo-3',5'-dichloro-4'-methoxybiphenylwas prepared likewise; the compound crystallized from petroleum ether (b.p. 60-68") in needles (yield 47Yc), m.p. 92". Anal. Found: C , 47.40; H , 2.57; Br, 24.15; CI, 21.22. 4-Acetamido-3',5'-dichloro-4'-methoxybiphenylwas prepared from the amine and acetic anhydride; after crystallization from ethanol, m.p. 218'. Anal. Calcd. for Cl5HI3NOCl2: S , 4.76. Found: N, 4.61. 4-Acetamido-3 ',5 '-dichloro-4'-methoxy-3-nitrobiphenyl.--Nitration of the above acetamidobiphenyl gave the nitro derivative in almost theoretical yield, crystallizing from acetic acid in yellow flakes, m.p. 213". Anal. Calcd. for C1SH12N203C12:N , 8.26. Found: N , 8.06. 4-Amino-3',5 '-dichloro-4'-methoxy-3-nitrobiphenyl .-Hydrolysis of the acetyl derivative gave the aminodichloromethoxynitrobiphenyl which separated from ethanol in bright red crystals (yield 99(&), m.p. 212", depressed by the acetyl derivative. Anal. Calcd. for C13H1~S203C1~: N , 8.95. Found: N, 9.11. 3,5-Dichloro-4-methoxy-3'-nitrobiphenyl.-The amine (2.0 g . ) was diazotized in dilute sulfuric acid and the diazonium solution added t o cold hypophosphorous acid (20 mi. of 305%). Next day the dichloromethoxynitrobiphenyl was isolated with ether and crystallized from ethanol, forming a yellow powder (1.89 g., 95%),

in Table I, together with the corresponding values for nitration with nitric acid in acetic anhydridei and for phenylation by phenyl radicals.6 TABLE I RELATIVE RATECOXSTANTS FOR OVER-ALL SVBSTITUTION OF BEXZENEDERIVATIVES B Y VARIOUS AGENTS

Results The relative rate constants for the over-all reactions of I b with the various substituted benzenes are shown

--Relative Ib

X in C6HdX-

rates of substitution by:---. HNOa AcrO Ph

+

H

a

1 0 15 033 .03

1 F 0.388 c1 429 Br 345 Me0 1.276 MeOOC 0.522 CN 0.335 \.slue for ethyl benzoate.

1 1 35 1.44 1 75 2 5

0.0037" 3.6

..

The proportions of isomers in the products from I b are shown in Table 11; note t h a t in the case of chlorobenzene or bromobenzene, one product was 2,B-dichloro-4-phenylanisole (VJII, X = H ) , formed by displacement of chlorine or bromine from the substrate. TABLE I1 PROPORTIONS OF PRODUCTS FROM REACTION OF Ib Proportions of products, 7'---mrla Unsubstd.a ortho

LVITH

PhS

7 -

X in P h X

para

Oh 37.8 62.1 Oh 32 62 c1 Oh 31.6 42.3 Br 27.25 72 8 Oh Me0 0 35.1 47.2 17 7 MeOOC CN 0 45.0 16.5 38 5 a I . e . , 2,6-dichloro-4-phenylanisole (VIII, X = H ) Less than 0.5% of the m-isomer could have been detected in each case

F

0 6 26 0

Table I11 compares the partial rate factors for substitution by I a in the various benzene derivatives with those for nitration in acetic anhydridei and for phenylation by phenyl radicals.8 TABLE I11 PARTIAL RATEFACTORS FOR SUBSTITUTION IS P h X VARIOUSAGENTS Compound

PhF

Position 0

m

P PhCl

0

m

P PhBr

0

m

P PhOMe

0

m

P

n1.p. 161'.

A n a l . Calcd. for CI3Hl9SO3Cl2:S , 4.70. Found: N, 4.60. 3-Amino-3 ',5 '-dichloro-4 '-methoxybiphenyl -Reduction of the nitro compound as before gave the amine which was isolated as its hydrochloride, needles from ethanol-ether, m p . 228'. A n d . Calcd. for CI3H1?NOCl3:S , 5.47. Found: N, 5.32 3,5-Dichloro-3'-cyano-4-methoxybiphenyl was prepared as before from the amine by a Sandmeyer reaction; m . p . 165-166'. Anal. Calcd. for C14H9SOC12: C, 60.46; H , 3.26; N,5.04. Found: C, 59.90; H , 3.03; N , 5.06.

V O l . 86

PhCOOMe

PhCN

0

0.72 0 0.88 0.85 0 0.88 0.59 0 0.88 2 . 78 0 2 09 0.55

r a t e factor for reaction with:HPiOa ArzO Pb

+

0.056

0 0 79 029 0009 137 033 0011 112 . .

,0026"

.49

,0079"

P

37 45 17

001"

0

P

--

2 19 1.5 1.23 2 6 1 12 1.19 2 09 1 8 1 83

. .

m

m a

-Partial Ib

BY

. .

. .

i I

Values for PhCOOEt

Reaction of 1,3,5trideuteriobenzene with I b can give either 3',5'-dideuterio- (XX) or 2',4',B'-trideuterio( 7 ) C K Ingold a n d hl S S m i t h , J C h c m SOL, 90.5 (19381: 51 1. Bird and C K . Ingold, i b i d . , 918 (1918) ( 8 ) I). H H e y , e l ai., ibid.,1971,2094 (19521, 44, 3412 ( 1 9 5 3 ) ; 794. 3352 (1951), 6 (1955), 1463 (1950)

BIPHENYLS F R O M BENZENE-1 , 4 - D I A Z O O X I D E DERIVATIVES

June 20, 1964 TABLE I\'

ISOTOPIC SPECIES I Y 1,3,5-TRIDEUTERIOBESZESE REACTION PRODUCT OF Ib WITH A MIXTCRE OF IT WITH SORMAL BEKZENE

PROPORTIONS OF A N D I N THE

Proportion o -f 0

Compound

1,3,5-Trideuteriobenzene Reaction product

(7')containing

1 25 832

various numbers deuterium atoms-1 2 3

0 05 0 7

2 6 7 9

96 1 8 1

2,6-dichloro-4-hydroxybiphenyl (XIX) . The reaction was carried out with a mixture of normal benzene and 1,3,5trideuteriobenzene ( 5 :1 w.,/w). Table IV shows the deuterium content6 of the 1,3,5-trideuteriobenzene, and of the mixture of methyl ethers formed by methylation of the reaction product.

Discussion The most salient feature of the results listed in Tables I1 and I11 is the complete absence of m-isomer in the products from anisole, chlorobenzene, or bromobenzene. Examination of synthetic mixtures showed t h a t 0.5yoof the m-isomer could easily have been detected in each case. This seems to rule out completely a radical mechanism for the reaction; no aryl radical could possibly show selectivity of so high an order as this. The reaction of I b indeed shows an even greater selectivity than does nitration with nitric acid in acetic acid; both chlorobenzene and bromobenzene give about 1yo of m-nitrohalobenzene under these conditions. It had been reported3 previously that I a attacks chlorobenzenes exclusively in the 0- and p-positions; however, the analytical procedure used in t h a t investigation was less accurate than the method adopted here and the earlier conclusions were consequently less definite. The radical mechanism3 for the biaryl synthesis suffers from the further defect of not explaining why some benzene derivatives give polymers with I whereas others give biphenyl derivatives. The distinction certainly does not depend on the electronic nature of the substituents; thus fluorobenzene, dimethylaniline, and methyl benzoate give biphenyl derivatives almost exclusively, whereas chlorobenzene and bromobenzene give only polymers. The only difference between the two classes of substrates seems to be that the ones giving polymers contain heavy atoms (Cl, Br). There seems to be only one way in which these facts can be explained. Decomposition of I can give four different types of product; a carbene ( X I ) , a zwitterion ( X I I ) , or a biradical which can exist in either a singlet ( X I I I ) or triplet (XIV) state. The three species differ in the distribution of electrons between the r - M O ' s and the in-plane u - A 0 of the carbon atom para to oxygen. In X I there are six r-electrons and two electrons in the PAO; in XI1 there are eight a-electrons and none in the u-.%O; in XI11 or XIV there are seven r-electrons and one electron in the CT-AO.

t

XI

XI1

XI11

t

XIV

2425

The electronic wave functions of X I and XI1 have the same total symmetry; consequently X I and XI1 are not chemically distinct species, but possible resonance structures of one single species. Presumably this will approximate closely to the neutral carbene structure X I . The wave function of XI11 differs in symmetry from X I or X I I ; i t is antisymmetric for reflection in the nodal plane, whereas those of X I and XI1 are symmetric. Hence XI11 is a chemically distinct species from X I or XI1 rather than a mere resonance structure; XI11 will exist in equilibrium with (XI ++ XI1)-although of course the interconversion of the two forms would be expected to be extremely fast. The triplet structure XIV is also a chemically distinct species; the conversion of XI11 to XIV may be quite slow since i t involves an intersystem crossing and it should be greatly facilitated by heavy atoms in the vicinity of the odd electron. The initial thermal decomposition of I should produce the intermediate initially in a singlet state which should approximate the carbene structure X I . lvow this is a vinylog of an acyl carbene, and carbenes carrying + E substituents (such as acyl) are little stabilized by resonance since the p-orbital of the divalent carbon atom is empty. Carbenes are the conjugate bases of carbonium ions and are likewise stabilized best by electron-releasing (-E, -I) substituents. For this reason carbenes carrying + E substituents are comparable with carbene itself in reactivity; carbethoxycarbene and cyanocarbeneg react with benzene to form derivatives of cycloheptatriene or norcaradiene. I t is therefore tempting to attribute the biphenyl synthesis to a reaction of this kind between X I and the aromatic substrate; in this case the product would be a curious spiran that would be expected to isomerize very easily to a derivative of biphenyl. The formation of polymers from I in solvents containing chlorine or bromine must then involve the free radical IV. If we are right in supposing t h a t X I reacts with benzene derivatives as a carbene, and that X I is in very rapid tautomeric equilibrium with the singlet biradical X I I I , we must suppose the polymerization to involve the triplet biradical XIV. The role of solvents containing chlorine or bromine is then to accelerate the formation of XIV from X I ; the ability of solvents containing heavy atoms to accelerate intersystem crossings of this kind has been well recognized since a classic investigation by Kasha. l o I t may a t first sight seem strange that the solvent should have so great an effect, given that the carbene itself contains two heavy atoms. These, however, are remote from the critical orbital (the in-plane PXO para to oxygen) which contains the electron whose spin is to be reversed. This electron could only reach the heavy atom if there were a strong resonance coupling between the a - A 0 and the a-MO's and couplings of this kind are known to be small. These conclusions have been supported by further studies of the effect of alcohol on the reaction between I and chlorobenzene. If the alcohol acts as a hydrogen carrier, as suggested p r e v i ~ u s l y ,then ~ in the first place there should be an exchange between the ahydrogen atoms of the alcohol and the hydroxyl group (9) 51. J S. Dewar and R P e t t i t , J Chrm Soc.. 2026 (1956I (IO) M. Kasha, Dircussiotrs F a r a d a y Soc , 9, 14 i l 9 5 0 ) . J C h e m P h y q ao, 7 1 (1952).

2426

MICHAEL J. S. DEWARAND K. NARAYANASWAMI

1 reaction coordinate. Fig. I.-Potential energy diagram for the reaction of XI with benzene: ( a ) rate of isomerization of XV slower than final proton (b) rate of final proton transfer slower than transfer (---); rate of isomerization ( - - - - -),

of the 4-hydroxybiphenyl formed in the reaction, and in the second place tertiary alcohols should fail to act as catalysts. When, however, we used a ,a-dideuterioethanol as catalyst in the reaction, no such exchange took place, for the infrared spectrum of the product showed i t to be free from deuterium while the n.m.r. spectrum of the recovered alcohol showed no a-hydrogen. Moreover i t has been found” that t-butyl alcohol is an active catalyst although it contains no a-hydrogen, while diphenylamine is not, although it contains a hydrogen atom that can very easily be extracted by radicals. The most likely explanation of the action of alcohols seems to be that they form strong hydrogen-bonded complexes with the carbene X I , but not with the biradical XIV. This seems reasonable for the carbene must in fact be represented as a hybrid of the unchanged structure X I and the zwitterionic structure XII, whereas there are no unexcited ionic structures for the biradical XIV. The oxygen atom in the latter should therefore be less negatively charged and so should form weaker hydrogen bonds. Hydrogen bonding may affect the relative rates of attack by X I on the solvent, and of isomerization of X I to XIV, in two ways. First, such solvation should increase the electrophilicity of X I and so increase the rate of attack on the solvent; second, the extra energy of solvation may conceivably make the carbene more stable than the biradical, or a t least reduce the rate of isomerization by making it less exothermic. Our conclusion then is that the biaryls are formed through attack by the carbene X I on the substrate; the observed orientation of the products moreover shows the over-all reaction to be an electrophilic substitution. We now have to explain why i t is that X I shows very high selectivity (greater than nitric acid in acetic acid) in choosing between the positions in a given (11) Unpublished work in this laboratory b y D r . J. Blackwell and b y LTrs N.1. Satelli.

Vol. 86

substrate molecule PhX, but very little selectivity in choosing between different substrate molecules PhX, PhY. This certainly runs entirely counter to Brown’s “selectivity rule”,’* a rule which can be shown by a simple MO treatment l 3 to follow from the conventional WhelandI4 mechanism for substitution. The only explanation seems t o be that X I reacts with PhX by a very exothermic reaction with very low activation energy to form an intermediate which can be written as a spiran (XV). The spiran can isomerize t o one or other of two benzenonium ions XVI, XVII, by rupture of one or the other of the bonds in the threemembered ring; these in turn form the final reaction products by shift of a proton from the substrate ring to oxygen. If the formation of XV has a very low activation energy, changes in the substrate will affect its rate but little. This can be seen very easily from an argumentI5 based on the use of potential curves, a type of argument first introduced by Bell, Evans, and Polanyi some time ago.16 It is easily shown that the rate of a highly exothermic reaction with a small activation energy should vary little with structural changes in the reactants. This general principle has been rediscovered from time to time by various authors; the original application’j was to explain the different products that may form when a mesomeric anion reacts with electrophilic agents of varying reactivity. I n the present case the argument is still clearer; for both the substrate PhX and the spiran XV are even alternant hydrocarbons, and so the heat of formation of XV from PhX should vary little with the nature of X , the resonance energy for the interaction between a substituent X and an even alternant hydrocarbon being small.” The heats of isomerization of XV to the benzenonium ions XVI or XVII will, however, differ greatly if X is a substituent with a selective stabilizing or destabilizing effect on carbonium ions”; if the isomerization has an appreciable activation energy, there may then be very significant rate differences for the conversion of XV into one or other of the possible isomeric benzenonium ions. At the same time the formation of XV may still be the rate-controlling step of the over-all process, provided that this reaction is so exothermic that the transition state for the isomerization of XV remains lower in energy than that for the formation of XV. These relations are illustrated in Fig. 1. The second step can have a large activation energy even though the first step is still rate determining. In these circumstances there may be large differences in rate for the formation of the various possible products from a given substrate PhX, but little between the rates of attack on different substrates PhX, PhY.

xv

XVI

XVII

(12) H. C . Brown and K . I, Nelson. J . A m . Chem. Soc., 7 6 , 8293 t1953), 77, 2300 (19.55). (13) M . J . S Dewar, T. Mole, a n d E . 1%’ T. U’arford, J . C h r m Soc., 3581 (1956) (14) G W . W h e l a n d , J A m Chem Soc., 64, 900 11942) (15) M J S D e w a r , Discussrons F a r a d a y Soc , 2 , 261 (1947) (16) Cf.M G E v a n s a n d M. Polanyi, T r a n s F a r a d a y S o c , 3 2 , 1340 (1936); R P Bell, P r o c R o y . Soc. ( L o n d o n ) , 8 1 5 4 , 414 (1938). (17) M. J. S Dewar, J . A m Chem Soc., 74, 3341, 334.5, 3350. 3353, 3355, 3357 rlY.52)

BIPHENYLS FROM BENZENE1 , 4 - D I A Z O O X I D E DERIVATIVES

June 20, 1964

One final point needs to be settled. The conversion of XV to the product involves two steps, isomerization to XVI or XVII, and loss of a proton from carbon to form the product. Either of these last two steps might be rate det,ermining; indeed, there is a third alternative, that ring opening and deprotonation occur synchronously by an EN^ mechanism. In order to distinguish between these mechanisms, we studied the reaction of I b with 1,3,5-trideuteriobenzene.Here the intermediate is XVIII, and the final product can be either X I X or X X . If the loss of the proton occurs in the slow step, there should be a primary isotope effect favoring the formation of X I X .

XVIII

.:w

XIX

c1

xx

D

.as a matter of convenience we carried out the experiment with a mixture of 1,3,5trideuteriobenzene and normal benzene; in this way we could get a free check on any possible isotope effect on the over-all reaction The results in Table I V show that benzene and trideuteriobenzene react a t essentially identical rates with X I , and that X I X and X X are formed in equal amounts. Table V compares the observed proportions of undeuterated, dideuterated, and trideuterated dichlorophenylhydroxybiphenyls in the reaction product with those calculated assuming zero isotope effects. The values agree within the limits of experimental error, showing that the product-determining step in the conversion of XV to the final product is the isomerization of XV to XVI or XVII. The course of the reaction is that indicated by the solid line in Fig. 1. TABLE V COMPARISOXS OF OBSERVED A N D PREDICTED PROPORTIONS OF ISOTOPIC SPECIES FROM THE REACTION OF Ib WITH BENZEXE CONTAINING 1 , 3 , 5 - T R I D E l J T E R I O B E S z E N E Deuterium atoms in product

Proportion of products obsd., % Proportion of products calcd. assuming zero isotopic effect

( 5 1 \V./W.) 0

2

3

83.8 8 . 0 8 . 2 84.0 8 . 0 8 . 0

Mechanism of Electrophilic Substitution.-The results discussed here have a bearing on the mechanism of electrophilic aromatic substitution, Some time ago one of usl8 suggested that the first step in such reactions might be the formation of a n-complex from the substrate and the electrophile, this in turn isomerizing to the benzenonium intermediate postulated by Wheland. If this is so, one can distinguish two cases. In the first, the n-complex is formed irreversibly in the 118) M. J. S. Dewar, "The Electronic Theory of Organic Chemistry," Clarendon Press, Oxford, 1949.

2427

rate-determining step. In the second, the a-complex is formed reversibly and the slow step is its isomerization to one or other of the possible benzenonium ion intermediates. One can see at once that only the first case is chemically significant; in the second, the formation of the a-complex can have no effect either on the oT-erall rate of reaction or on the proportions of ison,ers formed. Our biphenyl synthesis is an electrophilic substitution of the former type, where formation of the complex is rate determining. I t may seem at first sight that the spiran intermediate XV is far removed in structure from a n-complex, but this is not in fact the case. Just as the classical (XXI) and n-complex (XXII) forms of ethylene oxide are closely related, depending only on the state of hybridization of the carbon atoms, so also is XV related to an analogous n-complex in which the benzene ring essentially retains its integrity. Olah and his collaborator^^^ have observed similar relationships in the nitration of a variety of alkylbenzenes by the free nitronium ion, they have interpreted their results in terms of the a-complex mechanism. These reactions seem to follow the course indicated by the potential curves of Fig. l a (solid line), the formation of the n-complex being rate determining ; here again one can write the intermediate either as a n-complex (XX111) or with a three-membered ring (XXIV).

XXI

XXII

XXIII

XXIV

Our case is particularly striking in the complete lack of selectivity shown by XI to different benzene derivatives; anisole and methyl benzoate would undergo nitration a t rates differing by many orders of magnitude while their rates of reaction with X I differ less than threefold. Yet X I shows greater selectivity between the different positions in chlorobenzene than does nitric acid in acetic anhydride! These results suggest that the possibility that a-complex formation may be rate determining should always be kept in mind when studying reactions involving electrophilic substitution. Recent investigations by Kunitake and Pricez0 and by Stille, Cassidy, and PlummerZ1 support the general mechanism suggested above. These authors independently studied the thermal and photochemical decomposition of 3,5-dimethylbenzene-1,4-diazooxide in chlorobenzene,20 nitrobenzene, and tetrahydrofuran. *O 2 1 In the first case only biphenyl derivatives were formed, in the second nitrosobenzene, and 2,B-dimethylbenzoquinone (cf. 3), in the third a polymer. These reactions were interpreted in terms of an electrophilic attack on the solvent by an intermediate formulated as a carbene (XI, X = Me) or zwitterion ( X I I , X = Me), in complete agreement with the mechanism suggested here. (19) G . A . Olah, S. J . K u h n , and S. H Flood, J A m Chem S o c . , 83, 4571, 4 5 8 1 (1961) (20) T . Kunitake and C C. Price, i b i d . , 85, 761 (1963). (21) J. K Stille, P Cassidy, and L Plummer, ibid , 86, 1318 (1963)